Seven national freight origin-destination samples collected in Colombia from 1999 to 2005 were used to conduct an empirical investigation of the temporal stability of parameters of freight demand models in the short to medium term. Freight generation, freight distribution, and empty trip models were considered. To identify time-dependent effects, models were estimated with a panel formulation with time-dependent parameters and fixed time effects and then compared with the corresponding cross-sectional models. The results indicate the presence of statistically significant time-dependent effects on all freight generation models (production and attraction), freight distribution models (based on both loaded vehicle trips and commodity flows), and empty trip models. A literature review indicates that few studies are available on the temporal stability of parameters. The results show a remarkably consistent pattern in that the components of freight demand that could change faster (i.e., freight production and attraction) are those that exhibit the largest rates of parameter change. The rates of change for these models are 18.29% and 26.37%, respectively. In contrast, the freight distribution models of loaded trips were found to change less rapidly (10.50% and 1.94%, respectively, depending on the impedance function), while the tonnage distribution model exhibited only fixed time effects. The model that changes least rapidly is the empty trip model, which has a rate of change of 0.83%.